1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
|
/*
Genome-wide Efficient Mixed Model Association (GEMMA)
Copyright (C) 2011-2017, Xiang Zhou
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "gsl/gsl_matrix.h"
#include <cmath>
#include <iomanip>
#include <vector>
#include <cblas.h>
#include "debug.h"
#include "fastblas.h"
#include "mathfunc.h"
#ifndef NDEBUG
#include "eigenlib.h"
#endif
using namespace std;
/*
Reasonably fast function to copy data from standard C array into
gsl_matrix. Avoid it for performance critical sections.
*/
gsl_matrix *fast_copy(gsl_matrix *m, const double *mem) {
auto rows = m->size1;
auto cols = m->size2;
if (is_strict_mode()) { // slower correct version
for (auto r=0; r<rows; r++) {
for (auto c=0; c<cols; c++) {
gsl_matrix_set(m,r,c,mem[r*cols+c]);
}
}
} else { // faster goes by row
for (auto r=0; r<rows; r++) {
auto v = gsl_vector_calloc(cols);
assert(v->size == cols);
assert(v->block->size == cols);
assert(v->stride == 1);
memcpy(v->block->data,&mem[r*cols],cols*sizeof(double));
gsl_matrix_set_row(m,r,v);
}
}
return m;
}
/*
Helper function fast_cblas_dgemm runs the local dgemm
*/
void fast_cblas_dgemm(OPENBLAS_CONST enum CBLAS_ORDER Order,
OPENBLAS_CONST enum CBLAS_TRANSPOSE TransA,
OPENBLAS_CONST enum CBLAS_TRANSPOSE TransB,
OPENBLAS_CONST blasint M,
OPENBLAS_CONST blasint N,
OPENBLAS_CONST blasint K,
OPENBLAS_CONST double alpha,
OPENBLAS_CONST double *A,
OPENBLAS_CONST blasint lda,
OPENBLAS_CONST double *B,
OPENBLAS_CONST blasint ldb,
OPENBLAS_CONST double beta,
double *C,
OPENBLAS_CONST blasint ldc) {
#ifndef NDEBUG
size_t i,j;
if (is_debug_mode()) {
printf (" Top left corner of matrix A: \n");
for (i=0; i<min(M,6); i++) {
for (j=0; j<min(K,6); j++) {
printf ("%12.0f", A[j+i*K]);
}
printf ("\n");
}
printf ("\n Top left corner of matrix B: \n");
for (i=0; i<min(K,6); i++) {
for (j=0; j<min(N,6); j++) {
printf ("%12.0f", B[j+i*N]);
}
printf ("\n");
}
printf ("\n Top left corner of matrix C: \n");
for (i=0; i<min(M,6); i++) {
for (j=0; j<min(N,6); j++) {
printf ("%12.5G", C[j+i*N]);
}
printf ("\n");
}
cout << scientific << setprecision(3) << "* RowMajor " << Order << "\t" ;
cout << "transA " << TransA << "\t" ;
cout << "transB " << TransB << "\t" ;
cout << "m " << M << "\t" ;
cout << "n " << N << "\t" ;
cout << "k " << K << "\n" ;
cout << "* lda " << lda << "\t" ;
cout << "ldb " << ldb << "\t" ;
cout << "ldc " << ldc << "\t" ;
cout << "alpha " << alpha << "\t" ;
cout << "beta " << beta << "\n" ;
cout << "* A03 " << A[3] << "\t" ;
cout << "B03 " << B[3] << "\t" ;
cout << "C03 " << C[3] << "\t" ;
cout << "Asum " << sum(A,M,K) << "\t" ;
cout << "Bsum " << sum(B,K,N) << "\n" ;
cout << "Csum " << sum(C,M,N) << "\n" ;
}
#endif // NDEBUG
cblas_dgemm(Order,TransA,TransB,M,N,K,alpha,A,lda,B,ldb,beta,C,ldc);
#ifndef NDEBUG
if (is_debug_mode()) {
printf (" Top left corner of matrix A (cols=k %i, rows=m %i): \n",K,M);
for (i=0; i<min(M,6); i++) {
for (j=0; j<min(K,6); j++) {
printf ("%12.0f", A[j+i*K]);
}
printf ("\n");
}
printf ("\n Top left corner of matrix B: \n");
for (i=0; i<min(K,6); i++) {
for (j=0; j<min(N,6); j++) {
printf ("%12.0f", B[j+i*N]);
}
printf ("\n");
}
printf ("\n Top left corner of matrix C: \n");
for (i=0; i<min(M,6); i++) {
for (j=0; j<min(N,6); j++) {
printf ("%12.5G", C[j+i*N]);
}
printf ("\n");
}
}
#endif // NDEBUG
}
/*
Helper function fast_cblas_dgemm converts a GEMMA layout to cblas_dgemm.
*/
static void fast_cblas_dgemm(const char *TransA, const char *TransB, const double alpha,
const gsl_matrix *A, const gsl_matrix *B, const double beta,
gsl_matrix *C) {
// C++ is row-major
auto transA = (*TransA == 'N' || *TransA == 'n' ? CblasNoTrans : CblasTrans);
auto transB = (*TransB == 'N' || *TransB == 'n' ? CblasNoTrans : CblasTrans);
// A(m x k) * B(k x n) = C(m x n))
auto rowsA = A->size1;
auto colsA = A->size2;
blasint M = A->size1;
blasint K = B->size1;
assert(K == colsA);
blasint N = B->size2;
// cout << M << "," << N "," << K << endl;
// Layout = CblasRowMajor: Trans: K , NoTrans M
blasint lda = (transA==CblasNoTrans ? K : M );
blasint ldb = (transB==CblasNoTrans ? N : K );
blasint ldc = N;
fast_cblas_dgemm(CblasRowMajor, transA, transB, M, N, K, alpha,
/* A */ A->data,
/* lda */ lda,
/* B */ B->data,
/* ldb */ ldb,
/* beta */ beta,
/* C */ C->data, ldc);
}
/*
Use the fasted/supported way to call BLAS dgemm
*/
void fast_dgemm(const char *TransA, const char *TransB, const double alpha,
const gsl_matrix *A, const gsl_matrix *B, const double beta,
gsl_matrix *C) {
fast_cblas_dgemm(TransA,TransB,alpha,A,B,beta,C);
#ifndef NDEBUG
if (is_strict_mode() && !is_no_check_mode()) {
// ---- validate with original implementation
gsl_matrix *C1 = gsl_matrix_alloc(C->size1,C->size2);
eigenlib_dgemm(TransA,TransB,alpha,A,B,beta,C1);
enforce_msg(gsl_matrix_equal(C,C1),"dgemm outcomes are not equal for fast & eigenlib");
}
#endif
}
|
